Search

Highlights • This work presents the methodologies and evaluation results for the WHS algorithms selected from the submissions to the Multi-Modality Whole Heart Segmentation (MM-WHS) challenge, in conjunction with MICCAI 2017. • This work introduces the related information to the challenge, discusses the results from the conventional methods and deep learning-based algorithms, and provides insights to the future research. • The challenge provides a fair and intuitive comparison framework for methods developed and being developed for WHS. • The challenge provides the training datasets with manually delineated ground truths and evaluation for an ongoing development of MM-WHS algorithms., Graphical abstract This manuscript presents the methodologies and evaluation results for the WHS algorithms selected from the submissions to the Multi-Modality Whole Heart Segmentation (MMWHS) challenge, in conjunction with MICCAI-STACOM 2017. The challenge provides 120 three-dimensional cardiac images covering the whole heart, including 60 CT and 60 MRI volumes, all acquired in clinical environments with manual delineation. Ten algorithms for CT data and eleven algorithms for MRI data, submitted from twelve groups, have been evaluated. The results show that many of the deep learning (DL) based methods achieved high accuracy, even though the number of training datasets were limited. Several of them also reported poor results in the blinded evaluation, probably due to over fitting in their training. The conventional algorithms, mainly based on multi-atlas segmentation, demonstrated robust and stable performance, even though the accuracy is not as good as the best DL method in CT segmentation. The challenge, including provision of the annotated training data and the blinded evaluation for submitted algorithms on the test data, continues as an ongoing benchmarking resource., Knowledge of whole heart anatomy is a prerequisite for many clinical applications. Whole heart segmentation (WHS), which delineates substructures of the heart, can be very valuable for modeling and analysis of the anatomy and functions of the heart. However, automating this segmentation can be challenging due to the large variation of the heart shape, and different image qualities of the clinical data. To achieve this goal, an initial set of training data is generally needed for constructing priors or for training. Furthermore, it is difficult to perform comparisons between different methods, largely due to differences in the datasets and evaluation metrics used. This manuscript presents the methodologies and evaluation results for the WHS algorithms selected from the submissions to the Multi-Modality Whole Heart Segmentation (MM-WHS) challenge, in conjunction with MICCAI 2017. The challenge provided 120 three-dimensional cardiac images covering the whole heart, including 60 CT and 60 MRI volumes, all acquired in clinical environments with manual delineation. Ten algorithms for CT data and eleven algorithms for MRI data, submitted from twelve groups, have been evaluated. The results showed that the performance of CT WHS was generally better than that of MRI WHS. The segmentation of the substructures for different categories of patients could present different levels of challenge due to the difference in imaging and variations of heart shapes. The deep learning (DL)-based methods demonstrated great potential, though several of them reported poor results in the blinded evaluation. Their performance could vary greatly across different network structures and training strategies. The conventional algorithms, mainly based on multi-atlas segmentation, demonstrated good performance, though the accuracy and computational efficiency could be limited. The challenge, including provision of the annotated training data and the blinded evaluation for submitted algorithms on the test data, continues as an ongoing benchmarking resource via its homepage (www.sdspeople.fudan.edu.cn/zhuangxiahai/0/mmwhs/).

Unrepaired defects in the annulus fibrosus of intervertebral disks are associated with degeneration and persistent back pain. A clinical need exists for a disk repair strategy that can seal annular defects, be easily delivered during surgical procedures, and restore biomechanics with low risk of herniation. Multiple annulus repair strategies were developed using poly(trimethylene carbonate) scaffolds optimized for cell delivery, polyurethane membranes designed to prevent herniation, and fibrin-genipin adhesive tuned to annulus fibrosus shear properties. This three-part study evaluated repair strategies for biomechanical restoration, herniation risk and failure mode in torsion, bending and compression at physiological and hyper-physiological loads using a bovine injury model. Fibrin-genipin hydrogel restored some torsional stiffness, bending ROM and disk height loss, with negligible herniation risk and failure was observed histologically at the fibrin-genipin mid-substance following rigorous loading. Scaffold-based repairs partially restored biomechanics, but had high herniation risk even when stabilized with sutured membranes and failure was observed histologically at the interface between scaffold and fibrin-genipin adhesive. Fibrin-genipin was the simplest annulus fibrosus repair solution evaluated that involved an easily deliverable adhesive that filled irregularly-shaped annular defects and partially restored disk biomechanics with low herniation risk, suggesting further evaluation for disk repair may be warranted.Statement of significanceLower back pain is the leading cause of global disability and commonly caused by defects and failure of intervertebral disk tissues resulting in herniation and compression of adjacent nerves. Annulus fibrosus repair materials and techniques have not been successful due to the challenging mechanical and chemical microenvironment and the needs to restore biomechanical behaviors and promote healing with negligible herniation risk while being delivered during surgical procedures. This work addressed this challenging biomaterial and clinical problem using novel materials including an adhesive hydrogel, a scaffold capable of cell delivery, and a membrane to prevent herniation. Composite repair strategies were evaluated and optimized in quantitative three-part study that rigorously evaluated disk repair and provided a framework for evaluating alternate repair techniques. (C) 2015 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

International challenges have become the standard for validation of biomedical image analysis methods. Given their scientific impact, it is surprising that a critical analysis of common practices related to the organization of challenges has not yet been performed. In this paper, we present a comprehensive analysis of biomedical image analysis challenges conducted up to now. We demonstrate the importance of challenges and show that the lack of quality control has critical consequences. First, reproducibility and interpretation of the results is often hampered as only a fraction of relevant information is typically provided. Second, the rank of an algorithm is generally not robust to a number of variables such as the test data used for validation, the ranking scheme applied and the observers that make the reference annotations. To overcome these problems, we recommend best practice guidelines and define open research questions to be addressed in the future., Comment: Article published in Nature Communications: https://rdcu.be/bRmNr

Given the complex dynamics of cardiac motion, understanding the motion for both normal and pathological cases can aid in understanding how different pathological conditions effect, and are affected by cardiac motion. Naturally, different regions of the left ventricle of the heart move in different ways depending on the location, with significantly different dynamics between the septal and free wall, and basal and apical regions. Therefore, studying the motion at a regional level can provide further information towards identifying abnormal regions for example. The 4d left ventricular motion of a given case was characterised by a low number of parameters at a region level using a cardiac specific polyaffine motion model. The motion was then studied at a regional level by analysing the computed affine transformation matrix of each region. This was used to examine the regional evolution of normal and pathological subjects over the cardiac cycle. The method was tested on 15 healthy volunteers with 4d ground truth landmarks and 5 pathological patients, all candidates for cardiac resynchronisation therapy. Visually significant differences between normal and pathological subjects in terms of synchrony between the regions were obtained, which enables us to distinguish between healthy and unhealthy subjects. The results indicate that the method may be promising for analysing cardiac function.keywordscardiac magnetic resonancecardiac resynchronisation therapyheart failure patientleft bundle branch blockcardiac motionthese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

International audience; Given the observed abnormal motion dynamics of patients with heart conditions, quantifying cardiac motion in both normal and pathological cases can provide useful insights for therapy planning. In order to be able to analyse the motion over multiple subjects in a ro- bust manner, it is desirable to represent the motion by a low number of parameters. We propose a reduced order cardiac motion model, reduced in space through a polyaffine model, and reduced in time by statistical model order reduction. The method is applied to a data-set of synthetic cases with known ground truth to validate the accuracy of the left ven- tricular motion tracking, and to validate a patient-specific reduced-order motion model. Population-based statistics are computed on a set of 15 healthy volunteers to obtain separate spatial and temporal bases. Re- sults demonstrate that the reduced model can efficiently detect abnor- mal motion patterns and even allowed to retrospectively reveal abnormal unnoticed motion within the control subjects.

International audience; Mandible fractures are classified depending on their location. In clinical practice, locations are grouped into regions at different scales according to anatomical, functional and esthetic considerations. Implant design aims at defining the optimal implant for each patient. Emerging population-based techniques analyze the anatomical variability across a population and perform statistical analysis to identify an optimal set of implants. Current efforts are focused on finding clusters of patients with similar characteristics and designing one implant for each cluster. Ideally, the description of anatomical variability is directly connected to the clinical regions. This connection is what we present here, by introducing a new registration method that builds upon a tree of locally affine transformations that describes variability at different scales. We assess the accuracy of our method on 146 CT images of femurs. Two medical experts provide the ground truth by manually measuring six landmarks. We illustrate the clinical importance of our method by clustering 43 CT images of mandibles for implant design. The presented method does not require any application-specific input, which makes it attractive for the analysis of other multiscale anatomical structures. At the core of our new method lays the introduction of a new basis for stationary velocity fields. This basis has very close links to anatomical substructures. In the future, this method has the potential to discover the hidden and possibly sparse structure of the anatomy.

PMID 23286041; International audience; Locally affine (polyaffine) image registration methods capture intersubject non-linear deformations with a low number of parameters, while providing an intuitive interpretation for clinicians. Considering the mandible bone, anatomical shape differences can be found at different scales, e.g. left or right side, teeth, etc. Classically, sequential coarse to fine registration are used to handle multiscale deformations, instead we propose a simultaneous optimization of all scales. To avoid local minima we incorporate a prior on the polyaffine transformations. This kind of groupwise registration approach is natural in a polyaffine context, if we assume one configuration of regions that describes an entire group of images, with varying transformations for each region. In this paper, we reformulate polyaffine deformations in a generative statistical model, which enables us to incorporate deformation statistics as a prior in a Bayesian setting. We find optimal transformations by optimizing the maximum a posteriori probability. We assume that the polyaffine transformations follow a normal distribution with mean and concentration matrix. Parameters of the prior are estimated from an initial coarse to fine registration. Knowing the region structure, we develop a blockwise pseudoinverse to obtain the concentration matrix. To our knowledge, we are the first to introduce simultaneous multiscale optimization through groupwise polyaffine registration. We show results on 42 mandible CT images.

International audience; The 3D modeling of the proximal femur is a valuable diagnostic tool for orthopedic surgery planning. The use of computed tomography is the most prominent modality to visualize bones both in terms of resolution as well as in term of bone/tissue separation. Towards reducing the impact of radiation to the patient, low-dose X-ray imaging systems have been introduced while still providing partial views with rather low signal-to-noise ratio. In this paper, we focus on automating the 3D proximal femur reconstruction from simultaneously acquired 2D views. A deformable model represented by triangulated mesh surfaces extends to a linear sub-space describing the variations across individuals. Segmentation consists of inferring a global deformation of 3D model followed by a local adaption based on the most prominent combination of the sub-space parameters. The ba-sis of which relies on the minimization of a cost function based on the biplanar projection of this model. To this end, we employ an active region model that aims at optimizing the 3D model parameters such that projection of surface is attracted from edge potentials, while creating an optimal partition between the bone class and the remaining structures. The global parameters of the model and the local ones are optimized through a gradient-free approach. Promising results demonstrate the potentials of our method compared to a supervised reconstruction technique.

International audience; Non-linear image registration is an important tool in many areas of image analysis. For instance, in morphometric studies of a population of brains, free-form deformations between images are analyzed to describe the structural anatomical variability. Such a simple deformation model is justified by the absence of an easy expressible prior about the shape changes. Applying the same algorithms used in brain imaging to orthopedic images might not be optimal due to the difference in the underlying prior on the inter-subject deformations. In particular, using an un-informed deformation prior often leads to local minima far from the expected solution. To improve robustness and promote anatomically meaningful deformations, we propose a locally affine and geometry-aware registration algorithm that automatically adapts to the data. We build upon the log-domain demons algorithm and introduce a new type of OBBTree-based regularization in the registration with a natural multiscale structure. The regularization model is composed of a hierarchy of locally affine transformations via their logarithms. Experiments on mandibles show improved accuracy and robustness when used to initialize the demons, and even similar performance by direct comparison to the demons, with a significantly lower degree of freedom. This closes the gap between polyaffine and non-rigid registration and opens new ways to statistically analyze the registration results.

International audience; We propose to increment a statistical shape model with surrogate variables such as anatomical measurements and patient-related information, allowing conditioning the shape distribution to follow prescribed anatomical constraints. The method is applied to a shape model of the human femur, modeling the joint density of shape and anatomical parameters as a kernel density. Results show that it allows for a fast, intuitive and anatomically meaningful control on the shape deformations and an effective conditioning of the shape distribution, allowing the analysis of the remaining shape variability and relations between shape and anatomy. The approach can be further employed for initializing elastic registration methods such as Active Shape Models, improving their regularization term and reducing the search space for the optimization.